Author Topic: Professional Diaries #2 Taxation - 25 years of progress?  (Read 8766 times)

John Short

  • *****
  • Posts: 524
Professional Diaries #2 Taxation - 25 years of progress?
« on: October 27, 2015, 12:41:10 GMT »
It has been “suggested” to me to follow in Gord Evans’ footsteps and produce something similar on taxation to his diary pieces centred on integrated planning.  So here goes...........To simplify it for myself I will attempt to look at my experience of three areas of taxation and how I have seen them evolve on the ground since the mid 1980s in various countries: tax policy, tax administration and revenue forecasting.  Of course these areas of taxation are not as discrete as this but it is a useful starting point.

My first taste of taxation in a development context was not really from the perspective of taxation at all.  It was centred around trade policy and the dialogue Governments had with the World Bank on reducing the number of import tariffs and narrowing their range as part of the Structural Adjustment Credit conditionalities.  Effective Protection and Domestic Resource Cost Analysis were the backbone of the work.  However, it is also important to put all of this into the overall policy context of hugely overvalued exchange rates, foreign exchange allocation by administered means and import and export licensing.  Investment was also permission based through licenses and discretionary tax holidays were the norm to compensate for the overall policy distortions that were in place to manage and guide development.  Non project donor support was provided in the form of foreign exchange for balance of payment support which was used to import a predetermined basket of goods or from a country where the support was bilateral.  The vast majority of revenue came from imports so clearly any reduction in tariff rates could have a negative impact unless the exchange rate changed.  The domestic tax base typically was not robust to make up any shortfall.  The resultant local currency from the sale of the balance of payment support boosted the budget.  The IMF focused on managing deficits. Budgeting and public expenditure management as discussed on the PFMBoard hardly existed.

Uganda in 1990 under the guidance of Emmanuel Tumusiime-Mutibile (PS Planning and Secretary to the Treasury now Governor of the Bank of Uganda) took great steps in eliminating many of the policy distortions by removing trade licenses and introducing foreign exchange bureaus.  There had been a Private Sector Assessment carried out by the World Bank and ODA (now DFID) which had identified policy distortions as constraints to development.  The PS pre-empted discussions on the report by taking action as he had come to the same conclusion himself without needing the report!  What followed then was a period of adjustment of tariff reduction and the development of domestic taxation and the creation of the Uganda Revenue Authority in 1991.  Sierra Leone followed a similar path working with a World Bank team led by Pirouz Hamidian Rad and Mike Conteh and John Karimu from the Ministry of Finance.  Ethiopia, PNG, Zambia, Ghana, Nigeria, Kenya, Rwanda, Tanzania have all taken these steps.

The removal of policy distortions and particularly the dependence on high import tariffs led the way for the development of wider tax policy in most countries.  Expanding a limited sales tax to a VAT has been an almost universal development.  Addressing income tax at personal and company levels has also been a notable development.

Looking back as I am doing now and comparing the present to the situation it is quite staggering to see the changes in tax policy that have taken place.  However as Richard Bird has written “tax administration is tax policy” so the accompanying developments in tax administration cannot be ignored.  Tax administration will be looked at next.

If anyone is interested in readable articles on taxation I recommend Richard Bird’s articles which are available on the web.  I was fortunate to be in Belarus at the same time as him and hugely benefited from his wit and wisdom as well as discussions on taxation (over a beer or two).  It was 1993 so it was either that or the opera/ballet at the Bolshio Theater.  All three were sampled.
« Last Edit: October 14, 2016, 22:18:42 GMT by John Short »

John Short

  • *****
  • Posts: 524
Re: Professional Diaries #2 Taxation - 25 years of progress?
« Reply #1 on: November 08, 2015, 19:49:31 GMT »
Moving on to tax administration............

In addition to the maxim that tax administration is tax policy, there is a second quite relevant adage: the pursuit of the perfect is the enemy of the good.  If the administration is not up to the job, then taxes will not be collected.  If taxes are too complex, they cannot be administered efficiently.  A critical aspect of any tax review is to ensure that any recommendations for reform are consistent with the ability of the administration to carry them out.  In other words, even if a country accepts a set of recommendations regarding changes in policy, one also needs to ask whether the administration is in sufficiently good shape to meet the demands placed upon it.  If not, then effective tax policy suffers as a result.

The primary function of the tax system is to generate revenue; its first goal is to ensure that this function is discharged effectively.  A second goal is to raise the economic efficiency of the taxation system.  This requires that, for a given level of revenue, it interferes as little as possible in decisions made by firms (about production, trade and investment) and by households (about consumption and savings).  A third goal is to lift the tax burden off the poorest households and to ensure that actual tax structures are both horizontally and vertically equitable.  This is likely to be carried out with some degree of progression. 

Reform or redirection of the tax system is likely, as its main aim, to simplify tax policy and strengthen tax administration to achieve these objectives in the most efficient manner.

The design of the tax system is thus critical to achieving the objectives set out for taxation.  In this regard, it is perhaps easier to indicate what a "good tax" should not be:

•   It should not be complex, difficult and costly to administer.
•   The number of rates should not be many (they should be kept to a minimum)
•   It should not provide strong incentives for evasion so that enforcement costs are increased.  This mainly argues for avoiding excessively high rates, but would also encompass the relationship between ease of evasion and severity of penalties.
•   It should not introduce serious economic distortions.
•   It should not treat taxpaying individuals and businesses in similar circumstances differently - horizontal equity.
•   Its administration and enforcement should not be selective or skewed in favour of those with the wherewithal to defeat the system - elements of vertical and horizontal equity.
•   The system should not be heavily dependent on one or two particular taxes, as changes in the tax base will have serious implications for revenue collection. and
•   A “not-not”: the scope for granting or receiving exemptions should be minimised.

Avoiding all of these “nots” will ensure that the tax system will be elastic - that is responsiveness to growth in the economy.  As GDP becomes larger and particularly as the formal cash economy grows, the absolute amounts of revenue collected will increase.
   The "good" tax rules are complemented by the "good" tax administration rules.  The following are indicative of these rules :
•   organise the tax machinery to get the most returns by concentrating scarce resources on the major issues;
•   see that the law is properly drafted and codified and changed if necessary to support sound administration;
•   see that the administration is properly organised, staffed, and trained, and has sufficient numbers and the wherewithal to do their job;
•   locate the tax payers, place them on rolls (computerised where appropriate), and examine returns and audit them;
•   obtain relevant information from government departments and elsewhere and use it to back up the tax collection process;
•   ensure that taxes due are collected; and
•   ensure that appropriate penalties are set and are properly applied.

The combination of the "good" tax and the "good" tax administration provides the basis for an evaluation of the way taxation is conducted in any country.

There is also one important feature of a tax system that has to be added to the above: that of stability.  Unfortunately, a stable "bad" tax system is not desirable.  Stability can only be justified when the system has been adequately established and/or reformed.  Once the system has been through an appropriate and adequate reform process, then it should be left substantially unchanged.

My experiences in observing these principles in action and indeed in applying them are varied.  One of the most dramatic impacts on tax administration was when the Minister of Finance in Uganda removed his ability to grant discretionary exemptions in the early 1990s.  The substantial and daily queues outside his office disappeared and officials were freed up to do other work.  Any exemption that was not in the tax code had to be approved by Parliament so transparency was improved.  This applied to NGOs and when approved the expenditure equivalent to the tax paid was treated as a government contribution to the project.  Public expenditure prioritisation became a factor in decision making.  Similarly, the discontinuation of duty free shops in Kampala ended significant leakages even if it meant diplomats had to go to Entebbe!

Codifying legitimate exemptions in the tax laws was a principle that gained traction with improvements in tax policy. Doing away with tax holidays and reducing the level of high company tax rates and “earning” tax holidays through depreciation allowances and loss carry over freed up officials in the Investment Promotion Agencies to promote rather than vet applications for an investment licence that provided tax holidays.  Indonesia did this in the 1980s and many other countries followed suite in the 1990s onwards.  As part of the reforms import tariffs began to be seen as a tool for protection so items such as machinery became zero rated and General Sales Taxes and Excise Tax started to fill the revenue gap that this caused.  The introduction of VAT became a feature of most taxation regimes given the cascading nature (tax on tax) of existing sales tax regimes and the narrowness of their coverage – on imports and domestic manufacturing.  An early reform in Sierra Leone was to allow sales tax on inputs to be offset against sales tax on outputs for domestic manufacturers (a manufacturing level type VAT) with an increase in the nominal rate to make the change revenue neutral.  A “proper” VAT followed at a later date once the SLRA was firmly established.

Customs Departments were the source of most revenue and often the most leakages.  But broadening the tax base required strengthening the collection of domestic taxes – direct and indirect).  The solution in many countries was to create a Revenue Authority that merged Customs and Domestic Tax Administrations: Uganda, Tanzania, Zambia, Rwanda and Sierra Leone were some of the first to be set up.  The good tax administration rules were the basis of their development albeit at times slowly and unevenly.  Large Tax Payer Units were established, Tax Identification Numbers adopted, self-assessment introduced; audit and tax information departments created and computerisation adopted.  (Customs Departments had been typically the first users of computers through the use of ASYCUDA and adopted it though with some difficult at first due to the less than user friendly software but with the development of the successive versions and the associated hardware and internet, most tax administrations now use it successfully).   One of – if not – the most important developments in the creation of RAs was the recognition that professional management and staff are crucial to their success.  Tax administration is a career in its own right and does not depend on rent seeking opportunities.

Bilateral and multi lateral support for improvements in tax administration was triggered by the creation of Revenue Authorities.  I had the good fortune to work with David Child in carrying out progress reports on DFID’s support to URA and SLRA and benefited greatly from his experience in UK Customs and the development of the combined Customs and Revenue but also working with South Africa RA and URA amongst others.

PEFA in the 2005/2011 methodology has three indicators (nine dimension) dealing with tax administration which has been collapsed to two indicators (seven dimensions) though essentially covering the same content in the revised 2015 methodology.  These dimensions reflect the principles outlined above with respect to administration.  From my own experience of PEFA assessments, the tax administration indicators scores have improved over time which reflects the efforts in addressing and improving tax administration.  However one area that appears to be standing still is in tax arrears and often the reason for that is the lack of a legal basis for writing off uncollectable debts so the debt plus accumulating penalties and interest remains on the books.  Bullet point two above needs to be applied!

Tax policy formulation and tax administration are best carried out by separate bodies to ensure the necessary skills are in the correct institution and to avoid potential conflict of interest.  However, it is essential that they speak to each other and that tax policy must recognise the ability of the administration to implement.  Two examples illustrate this important point.  In The Gambia, I was tasked with assisting in the creation of the Development Act.  The team developed an incentive structure that encouraged backward linkages to encourage hotels to source supplies from within the country (Gambia is a tourist country).  On returning years later I was told that the part of the Act had to be rescinded as the tax authorities could not administer it as companies were abusing the provisions by falsifying documentation and generally using the provisions as loopholes.  Nothing changes it would seem!  The second example was more positive.  In Zambia the design of a duty drawback system was developed alongside the ZRA and used the existing VAT refund procedures to implement the drawback.

Lessons?  Keep tax policy simple, get rid of discretionary exemptions and eliminate loopholes!  Apply the good tax principles.  Oh, if life was so easy......................................................!

postscript: for those that can access it see
or his article in print in The Times 9  Nov 2015 railing again complexity
« Last Edit: November 09, 2015, 20:12:54 GMT by John Short »

John Short

  • *****
  • Posts: 524
Re: Professional Diaries #2 Taxation - 25 years of progress?
« Reply #2 on: March 20, 2016, 21:40:14 GMT »
For the last of my Diary entries, I am going to steal or repeat what I have posted elsewhere on the PFMBoard and add a little.

Tax to GDP ratios differ quite dramatically often between countries with at what looks at first sight similar tax structures.  Often this difference is ascribed to tax policy (including the scope and extent of exemptions) and tax administration, while not fully looking at the main tax driver, the economy itself.  Indeed good tax policy and good tax administration should not distinguish between sectors in an economy with respect to how they are taxed.  Perhaps, then the most logical explanation of this is the observation that these higher tax-to-GDP-ratio countries have a higher share of monetised economy in GDP through formal activity in industry, agriculture and tourism.  Money means spending on taxed goods.  Earning money leads to payment of direct taxes. Increased income means spending on excisable goods.  The issue is not necessarily the composition of GDP but understanding how GDP is estimated.

Over the years I have modelled this hypothesis in work on revenue forecasting in Uganda, Rwanda and Tanzania and more recently in Albania. In Albania this was done in two time periods which has allowed observation on the evolution of the estimation of GDP.  The construction of Albania’s GDP is made up two main elements: the observed value added and the non observed value added.  In addition, agriculture has an own consumption element in its construction.  It can be assumed that non observed value added is what could be termed the informal sector and own consumption the non monetised sector. The observed sectors is the formal sector. The relative sizes of each of these elements will impact on the Albania’s tax base relative to GDP.  Estimating GDP by value added by sector in 2010, 39% of agriculture value added is not sold i.e. is not monetised and 17.5% of value added in all sectors minus public administration is generated using the non observed technique.  In 2004, comparable figures were 44% and 29%.   This would suggest that in 2010, 7.9% of GDP by sector value added is non monetised, 13.7% is informal, 64.5% is formal with a further 14% in public administration.  In 2004 the comparable figures were 11% of GDP is non monetised, 23% is informal, 58% is formal and 9% in public administration. This indicates that Albania’s tax base has expanded though still has room to grow further.

The two reports are attached.

This construction is useful in an attempt to explain and forecast revenue, as the contribution of the formal monetised sector to the tax revenue is likely to be higher than the informal sector’s. The informal sector is unlikely to be registered for VAT so its value added will not be taxed (its inputs, however, are likely to be, if purchased from the formal sector). The surplus of the informal sector may be taxed through presumptive taxes if the authorities have been able to register the entities, whereas formal sector companies will be taxed on their profits.  Formal sector companies' employees will be subject to PAYE, while informal entities will not.  The non monetised element of GDP will be outside the tax base completely.

How the composition of GDP changes can have an impact on revenue without any changes in tax policy and tax administration, though in reality improvements in the latter two may well influence the structure of GDP thought improving incentives and compliance.  Since 2007 Albania has embarked on tax policy and tax administration reform that will have improved the incentive structure and these will have yielded some improvements in the tax to GDP ratio  Tax reform is seen as a spur to investment to generate the growth needed for structural change.  However, significant improvements in the tax to GDP ratio will only be realised when the monetised formal economy has a significantly higher share of GDP than it presently has.  Indeed since 2007, tax to GDP ratios have remained relatively constant while nominal tax rates have decreased for many taxes which suggests that the improved tax administration and higher share of monetised and formal activity in GDP has compensated for the impact of lower rates.

It is possible to formulate a simple model based on the tax to GDP relationships and the structure of GDP to show what would happen to the tax to GDP ratio given changes in the structure of the economy.  Three basic relationships are modelled.
1. GDP grows annually - the various sectoral components can grow at a uniform rate or by different amounts.
2. With developments in the economy, there is a graduation from the non-monetised sector to the informal sector, and from the informal sector to the formal sector.
3. As the size of the formal sector grows relative to the informal sector and the informal sector grows relative to the non-monetised, there is a greater contribution to tax per unit of GDP.

Sizable increases in revenue to GDP ratios from one year to another will only happen if the composition – the structure - of GDP changes in the manner addressed here.

The PEFA Secretariat has produced a useful summary of tax models in Refinements for shadow testing PEFA 2015 TESTING VERSION ASSESSMENT TEAMS which I am taking the liberty to reproduce here as I suspect the audience may have not been significant.  This is making the heroic assumption that the audience is greater here!!!

Models for forecasting tax revenues

For forecasting tax revenues, mainly three basic types of models are used: Macroeconomic or GDP-based model, Micro-simulation model and Revenue receipts model.

Macroeconomic or GDP-based modelling requires the construction of data series of tax revenues and their bases for each type of tax, such as gross domestic product (GDP), consumption or imports. Using the two data series, the underlying relationship between the tax base and the tax revenue, known as tax elasticity, is estimated. The tax elasticity measures the “natural” response of the tax revenue with respect to changes in the tax base without any discretionary changes . It is essentially the ratio between the percentage change in tax revenue divided by the percentage change in the tax base and gives a measure of how tax revenues would change as a result of change in tax base only even when the tax rates remain unchanged.

Tax elasticity can be computed from year to year and then an average can be taken. Alternatively, if a decent time series for 10-12 years is available, a regression analysis would yield better estimate of tax elasticity. Once the elasticity is known, the future changes in tax revenues are nothing but the future changes in the base multiplied by the elasticity. So now the future values of tax base or changes in tax base have to be forecast using the estimate of GDP growth and the price indices in the future. For example, if import revenue is to be forecast in terms of aggregate goods imported and the effective import duty rate, then aggregate goods imports need to be forecast in terms of GDP and real import prices. This model can be applied to all types of tax revenues - VAT, Excises, income taxes and trade taxes with equal ease.

Micro-simulation models are based on forecasting tax revenues with the help of tax returns of individual taxpayers or transactions and then aggregating the results. The model contains a tax calculator that applies all the tax rules to the information in each tax return and then aggregates the tax liabilities across all the returns. By modifying the logic in the calculator, impact of alternative policies can be estimated. 

The main advantage of micro-simulation modelling lies in its capacity to estimate the distributional effect of a given policy proposal on particular groups of people or sectors of the economy. These models are capable of providing distributional impact analysis by identifying potential winners and losers in the society from proposed policy changes and thus they help assess the effects of a variety of policy options. The model can be also used to forecast tax revenues if the information in each tax return can be projected into future years on the basis of assumptions about economic growth, inflation rate, exchange rate etc. These models are best suited for income taxes and trade taxes. A variation of the model can also be applied to VAT revenues if the tax returns from individual vendors are available. This model can be used only when the data base is computerized.   

Revenue receipts model is a simple tool to monitor and project short-term revenues from major taxes, components of a tax, or non-tax charges.  It requires primarily monthly tax collection data, which are easily available from the tax agency or the Ministry of Finance. While this model is mainly targeted at short-term forecasting and for monitoring of monthly tax receipts over a financial year, it contains similar features to a macro-model for forecasting one-year or medium-term revenues as long as no major changes are expected in the tax or economic structure over the forecast period.

One simple approach in this model is to take the annual forecast of revenue collections of a specific tax and then distribute these collections over each month of the year on the basis of the seasonal patterns in the past. A more sophisticated model forecasts each month’s revenues on the basis of the corresponding month in the last year but adjusted for the expected growth rate in revenue collections due to real economic growth, inflation rate and tax rate changes. In any given month, annual tax receipts for the fiscal year can be expressed as the sum of two parts: (1) actual revenues collected up to the month for which receipts data are available; and (2) forecasted receipts for each of the remaining months of the fiscal year.  To project the second part of monthly receipts, the model takes into account the actual growth of the year-to-date tax collections as compared with that of the same period in the previous fiscal year, and the projected growth of tax base proxies (e.g. GDP, private consumption, imports, etc.).

VAT revenue forecasting

For forecasting of VAT revenues, in addition to the three standard models outlined earlier, two additional models specifically suited to VAT are available, thus in all five approaches may be adopted.  The two models are referred to as an aggregate or national accounts model and a disaggregate model, which is also referred to as Input-Output table model. The aggregate method starts with the GDP and arrives at the value-added included in the VAT base by adjusting the GDP for imports, zero rating, exemptions, and turnover threshold. The disaggregate method requires detailed breakdown of the National Accounts and Input-Output (I-O) Tables, and also information from other sources, such as household expenditure and industrial surveys to estimate the impact of exemptions and zero rating.  The tax base is computed by adding up the value added of each sector in the economy.

I have found the more "sophisticated" approach of the revenue receipts model useful as monthly forecasts can be replaced by actuals while preserving the integrity of the seasonality of the data and factoring in growth rates for each tax based on the individual drivers. Such a model for Palestine was produced under DFID PGF project.

Finally, one thing that the PEFA 2016 stresses in assessing forecasting techniques is that assumptions should be clearly set out.  A good recommendation.
« Last Edit: March 21, 2016, 11:30:21 GMT by John Short »


  • *
  • Posts: 154
Re: Professional Diaries #2 Taxation - 25 years of progress?
« Reply #3 on: March 29, 2016, 12:36:22 GMT »
An excellent presentation that clarifies the substance and appropriate strategy for a complex system. FF


  • *****
  • Posts: 348
Re: Professional Diaries #2 Taxation - 25 years of progress?
« Reply #4 on: March 30, 2016, 09:19:58 GMT »
I find the last post on approaches to forecasting particularly useful. Even if we've seen some of it before, it's handy to have this all in the same place. Thanks!

John Short

  • *****
  • Posts: 524
Re: Professional Diaries #2 Taxation - 25 years of progress?
« Reply #5 on: August 24, 2016, 09:32:20 GMT »
"I have found the more "sophisticated" approach of the revenue receipts model useful as monthly forecasts can be replaced by actuals while preserving the integrity of the seasonality of the data and factoring in growth rates for each tax based on the individual drivers. Such a model for Palestine was produced under DFID PGF project."

This was done in February 2015.  I have tested its robustness by comparing the 2015 actuals against the forecast for the 2015 baseline - the actual  total was 102% of the forecast total (with variations on individual taxes).  Inputting the actuals for months January to August (which simulate an update in September for budget preparation purposes), this was reduced to 101%.  Of course the test will also be how will the model do for 2016 and 2017?  But the model can be recalibrated by addressing the validity of the assumptions as well as establishing a new baseline using the most recent actual collections.


RSS | Mobile

© 2002-2023
Powered by SMF